Deep Learning For Damage Prognostic on Aircraft Engines

Published Apr 15, 20227 min read0 comments

This project was about using deep learning for damage prognostic on aircraft engines. The project was part of a computer science course called artificial intelligence and deep learning that I took for my master degree in computer science and informatics at Roskilde University.

The goal with the project was to find out when aircraft engines are about to break, so they can get repaired in time before further breakdown to reduce cost.

I used deep learning to provide insight into future prediction by looking at data from the past. The data set used in this project consisted of Turbofan Engine Degradation Simulation Data Set from Nasa, that included time-series data from simulated aircraft engines. The images below show some of the sensor readings from the dataset.

data visualization for aircraft sensor readings

I explored different deep learning architecture for this project. I worked on a simple linear regression model, and then explored multilayer perceptrons and recurrent neural networks (RNN) and long short term memory (LSTM).

The project was coded in Python with TensorFlow and Keras high level API.